The application of deep learning in communication signal modulation recognition

Automated Modulation Classification (AMC) has been applied in various emerging areas such as cognitive radio (CR). We also notice that Deep Learning (DL) is a powerful classification tool that has gained great popularity in various field. This article focuses on DL and aims at using it to solve communications problems. We propose a new data conversion algorithm in order to gain a better classification accuracy of communication signal modulation. This paper will show that our new method will bring significant improvement in signal modulation classification accuracy. Besides, AlexNet and GoogLeNet, two well-known DL network models, ResNet and VGG, will be utilized in this task to compare with each other.